Calling peaks from NRSF ChIP-Seq with USeq

[sonja]

Hello!

I am having some format troubles with the NRSF dataset downloaded from here:

http://www.wip.ncbi.nlm.nih.gov/projects/geo/query/acc.cgi?acc=GSE13047

The files are in txt format like:

Uniq files: all tags has single alignments in the genome
Column 1: tag sequence
Column 2: alignment score
Column 3: # of hit in the genome, 1 = unique hit
Column 4: Chr position
Column 5: Chr direction
Column 6: matched genome sequence
Column 7: next best possible alignment score


example:
head -5 GSM327023_chipFC1592_uniq_hg17.txt
GCAGAGTAACCCGCCCCACCCCACC 10406 1 chr6:156964520 F GCAGAGTAACTCTCCCCACCCCACC 9359


Now I want to run Useq on it. As I have to run the program a lot of times I would like to make it run with one command (ChIPSeq application) which supports ELAND format. So I was trying to convert my files into eland.

I was wondering if s.o. can help me here. Apparently I didn't manage a correct conversion. I don't seem to have produced the right amount of columns.
So my question is: How does the ELAND format read by Useq look like?

That's a line of the error meassage:

Error: line does not contain enough columns -> GCGCCGAGCATTCCGGCCTGAGGAG CTTCCCAGGCCGGAATGCTCGGCGC U1 1 0 0 chr2.fa 55756496 R .

If anyone could help here that would be great! As this NRSF dataset is widely used I might not be the only one having these troubles...
Thanks a lot!
dani

[ETHANol]

I know USeq accepts ELAND export format.

http://seqanswers.com/forums/showthread.php?t=1096

I don't think the format you posted is a standard ELAND format or maybe an older format. Your best bet would probably be to convert it to a BED file and use that.

I use this command to convert my ELAND files into QuEST files. It should be pretty straight forward to alter it to convert your files into a BED file.

I'd try this. No guarantee it will work. Load onto Galaxy and convert the colon to a tab with the "convert delimiters to tab" function. Export from galaxy and try.


For your file I'd try this (I use something similar to convert my ELAND files into QuEST files - disclaimer I'm not really sure what it does):

cat yourfile.txt | awk '{if($6 == "F"){print $4" "$5" +"}else{print $4" "$5+25" -"}}' | awk 'BEGIN{FS="[. ]"}{print $1" "$3" "$4}' > yourfile.bed

Then load back onto Galaxy and convert spaces to tabs as you did before.

Yeah, not the most straight froward approach but it should work.

Publicly available data is a real mess and methodology is poorly documented. Something the community should work on improving.

[sonja]

Hi,
thanks for your answer. But bed is not accepted by the USeq ChipSeq application.
About your way to convert to bed: if you replace " " by "\t" you don't have to use GALAXY.

Thanks anyway!

[ETHANol]

With USeq I think it's better to just put the data through the modules separately. It doesn't take too much more time. The ChipSeq wrapper gets hung up too easily.

Thanks for the tip.